This week in AI and ML news: An AI-centered class action lawsuit, Google’s 3D flyovers, and more.
Author’s Note
Business leaders agree that AI-powered solutions have a vital role to play in the ongoing fight against climate change and its myriad effects. Plainsight’s latest blog for IoTPlaybook looks at some of the ways organizations are deploying computer vision solutions to predict, track, and manage the effects of climate change around the globe as well as some of the obstacles still standing in the way.
AI News
The Lawsuit That Could Change AI Copyright Laws
Last June, Microsoft subsidiary GitHub introduced a coding assistant called GitHub Copilot. The AI-powered tool is trained on publicly available code and helps automate portions of the writing process for data scientists. More than a million developers have used the tool in its first year-plus, but plaintiffs in a new class action lawsuit say the tool’s unlicensed use of copyrighted code represents “software piracy on an unprecedented scale.”
Matthew Butterick, both a lawyer and a software developer, filed the lawsuit last Friday with the support of a San Francisco-based firm. In a statement, he notes the case’s potential to establish new precedents and change the world of AI copyright. “As far as we know,” Butterick says, “this is the first class-action case in the US challenging the training and output of AI systems . . . it will not be the last.”
Writing for, The Verge, Jamese Vincent clarifies that Microsoft and its partners are among the many tech giants to leverage copyrighted data in training solutions. Organizations have tended to claim fair use in defending their practices. Read Vincent’s interview with Butterick and two of his lawyers to learn more about the ongoing early stages of what could prove a landmark legal battle.
Can AI Set a Tasty Thanksgiving Table?
AI is capable of doing incredible things in the kitchen. Advanced solutions are helping restaurants maintain order accuracy, manage periods of high demand, and generate insights to provide better customer service. But what about recipes? Can AI-enhanced solutions suggest appetizing dishes or will digital chefs leave a bad taste in our mouths?
Four New York Times cooking columnists recently put AI’s recipe-writing powers to the test. Leveraging a neural network called GPT-3, Priya Krishna entered some personal and preferential info before asking the network to suggest a Thanksgiving menu tailored to her. She noted, for example, her Indian-American heritage and a distaste for cloying desserts. GPT-3, which is trained on texts including cookbooks, suggested dishes included pumpkin spice chaat, naan stuffing, and “cranberry sauce that’s not too sweet and a little spiced.”
Cooking the recipes proved occasionally confusing and tasting the finished products was mostly disappointing. There were few instructions related to sensory clues like scent and recipes often called for too many or too few ingredients. The recipe for turkey called for a 12-pound bird, one garlic clove, and no additional oil. “We’re not out of a job,” said Melissa Clark, another cooking columnist.
Watch a walkthrough and interview from the New York Times and listen to Krishna discuss her experience on CBC Radio’s As It Happens podcast.
Turning Still Images into 3D Nature Scenes
Research conducted by scientists at Google, UC Berkeley, and Cornell University could help people everywhere take to the skies and enjoy immersive natural experiences. Infinite Nature turns visual data into detailed, realistic scenes and may have applications in the gaming and augmented reality industry.
The central challenge of creating these 3D flyovers comes down to what Google calls perpetual view generation. This involves using a single input to create navigable, photorealistic 3D scenes. Initially, researchers addressed this challenge primarily with video, but challenges including the limited availability of high-quality data slowed their efforts. A new version of the solution, called InfiniteNature-Zero, was just unveiled at a conference last month and primarily uses landscape photos. Check out a summary of the ongoing research to learn more how the solution has evolved and what the future could hold.
About the Author & Plainsight
Bennett Glace is a B2B technology content writer and cinephile from Philadelphia. He helps Plainsight in its mission to make vision AI success repeatable, scalable and traceable for enterprises across industries.
Plainsight provides the unique combination of AI strategy, a visual data science platform, and deep learning expertise to develop, implement, and oversee transformative computer vision solutions for enterprises. From solution-centric strategy, through model deployment, and ongoing monitoring and oversight, Plainsight empowers enterprises to create and operationalize responsible vision AI applications for solving high business challenges.
